Official AMD Ryzen Benchmarks, Reviews, Prices, and Discussion

Page 140 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

Shmee

Memory & Storage, Graphics Cards Mod Elite Member
Super Moderator
Sep 13, 2008
7,382
2,419
146
Hmm, if ryzen really scales with memory frequency, then I am glad I got some fast DDR4. I know with X99 it makes little difference; my 2667 MHz is fine, especially with quad channel, but now I am glad I got 3200 MHz and I wonder if even faster would help.
 
  • Like
Reactions: looncraz

PotatoWithEarsOnSide

Senior member
Feb 23, 2017
664
701
106
If performance is scaling with memory bandwidth, surely the APUs due with/after Vega will perform admirably. They'll be coming with HBM2 @ 512GB/s, at least at the high end segment.
 

french toast

Senior member
Feb 22, 2017
988
825
136
If performance is scaling with memory bandwidth, surely the APUs due with/after Vega will perform admirably. They'll be coming with HBM2 @ 512GB/s, at least at the high end segment.
If you are speaking of raven ridge then you will be disappointed im afraid, likely not to have hbm, if it did i doubt we would see 200gb/s, it doesn't need it and would be wasting power.
We can dream though :)
 
Jan 15, 2017
39
54
61
And geekbench has special memory subtests that factor into final score.

This is basically a non-article.

Sampsa tested on Arma 3:

gv1um51.png


Thats average fps in there.

https://www.io-tech.fi/artikkelit/ylikellotustesti-ilmalla-ryzen-7-1800x-1700x/#4
 
  • Like
Reactions: lightmanek and ZGR

PotatoWithEarsOnSide

Senior member
Feb 23, 2017
664
701
106
I agree. I was merely looking at removing the memory bandwidth bottleneck that ultimately seems to exist on Ryzen CPUs. That is the correct conclusion from performance that scales with memory frequency, right?

IMO, needing the fastest memory to overcome another problem isn't ideal from a consumer standpoint. After all, sorting out the L3 issue should mitigate the need for faster memory...to a large extent. It'd be poor form to pass the solution over to the consumer's wallet.
 
  • Like
Reactions: french toast

french toast

Senior member
Feb 22, 2017
988
825
136
If you are speaking of raven ridge then you will be disappointed im afraid, likely not to have hbm, if it did i doubt we would see 200gb/s, it doesn't need it and would be wasting power.
We can dream though :)
People dont seem to realise AMD apus outperformed intel in gaming whilst being severely bottlenecked by excavator as well as memory bandwidth starved with either single channel and/ or low speed ddr3.
Ignoring vega graphics block on 14nm, just using 4/8 ryzen @ 2.3 base? 2.7 turbo 3.4ghz ST turbo as well as DDR 4 3200 will massively speed up gaming performance, any vega improvements will be on top of that.
HBM is likely too expensive right now for amds r&d budget.
 

lolfail9001

Golden Member
Sep 9, 2016
1,056
353
96
Sampsa tested on Arma 3:
Not bad improvement from just ~2ns of reduced memory latency (granted, 20% of bandwidth), actually. Granted, L3 interconnect works like 20% faster too.
People dont seem to realise AMD apus outperformed intel in gaming whilst being severely bottlenecked by excavator as well as memory bandwidth starved with either single channel and/ or low speed ddr3.
What CPU bottlenecks are you talking about with FREAKING APUs? AMD APUs outperformed Intel's iGPUs in gaming until Skylake/Broadwell-C simply because they were faster. Once memory became the bottleneck, they evened out around GT9800 performance.
After all, sorting out the L3 issue should mitigate the need for faster memory...to a large extent.
I mean, L3 issue is interconnected with memory speed, since the interconnect runs at basically memory clock.
 

french toast

Senior member
Feb 22, 2017
988
825
136
Not bad improvement from just ~2ns of reduced memory latency (granted, 20% of bandwidth), actually. Granted, L3 interconnect works like 20% faster too.

What CPU bottlenecks are you talking about with FREAKING APUs? AMD APUs outperformed Intel's iGPUs in gaming until Skylake/Broadwell-C simply because they were faster. Once memory became the bottleneck, they evened out around GT9800 performance.

I mean, L3 issue is interconnected with memory speed, since the interconnect runs at basically memory clock.
There is no question in cpu intensive titles intel apus got a large boost in comparison to amd, sometimes overtaking them.
Admittedly i did not see the latest round of bristol ridge vs kabylake 15w-35w tests, but i imagine it cant be any different.
Intel usually were put in higher end laptops so benefited from dual channel always, as well as higher spec memory so finding comparison is difficult.
I do know in some instances skylake out performed carizzo in certain laptop configurations.
 

flash-gordon

Member
May 3, 2014
123
34
101
If you are speaking of raven ridge then you will be disappointed im afraid, likely not to have hbm, if it did i doubt we would see 200gb/s, it doesn't need it and would be wasting power.
We can dream though :)
I believe HBM on APUs are now more probably than not. The IPs are already there, HBC is working, Ryzen performance is real, the stars are aligned...
 

Glo.

Diamond Member
Apr 25, 2015
5,662
4,421
136
I have changed my mind twice before about APUs, and maybe it is time to change it again.

What is the reason you guys believe that Raven Ridge APUs will NOT have HBM2? Apart from price?

At this very moment, just based on technicalities of architectures, I see more reasons to use the HBM2 on APU package, than not to use.

The only reason why not to use HBM2 is, maybe price. Overal production costs for Raven Ridge APU with HBM2 can be lower than similarly die sized GPU with GDDR5.
 
  • Like
Reactions: Drazick

french toast

Senior member
Feb 22, 2017
988
825
136
I don't really it's a question if Ryzen scales with memory bandwidth, and more like if game scales with memory bandwidth. It's same with Intel's cpu's.

Here is Memory scaling on 1800x in Arma 3.

ryzen-bench-oc-mem-arma3.png


source:

https://www.io-tech.fi/artikkelit/ylikellotustesti-ilmalla-ryzen-7-1800x-1700x/
Its not just the cpu scaling in apus with memory bandwidth, its more the gpu, thats the main issue alongside thermal headroom imo.
Raven Ridge stands to get a large multiple boosts to its performance in everything but especially gaming.

Ryzens uarch and 14nm lpp (LPU for RR? ) excel at frequency below 3.5ghz, ryzens ST performance dwarfs excavator, which alongside the double up in threads is going to remove any bottleneck that existed in certain titles with low power excavator.(majority gpu bound, but a boost likely in all titles imo)
Raven Ridge supports faster DDR4 which alongside the better IMC will boost bandwidth to the whole APU.
Vega is going to be significantly faster than GCN3, both in IPC, frequency, also bandwidth efficiency.
Raven Ridge will be home baked on either 14nm LPP or 14nm LPU (wildcard), which will give significantly more thermal headroom for the APU, although personally if feel lpp is more suitable /likely with any move to a HP version os samsungs 14nm likely to come with zen+ (if at all)
 

french toast

Senior member
Feb 22, 2017
988
825
136
I believe HBM on APUs are now more probably than not. The IPs are already there, HBC is working, Ryzen performance is real, the stars are aligned...
well like to believe it wont, that way i cant fet disappointed :)
besides intel Wouldn't be happy with that would they :)
 

french toast

Senior member
Feb 22, 2017
988
825
136
One more thing. If Ryzen 5 is both 6C and 4C lineups, that means that Ryzen 3 are going to be APUs.
Correct, timing makes sense.

Edit, HBM is coming no doubt, but judging by some statemens from amd it seems unlikely this year.
Plus i dont like getying my hopes inflated :)
 

lolfail9001

Golden Member
Sep 9, 2016
1,056
353
96
What is the reason you guys believe that Raven Ridge APUs will NOT have HBM2? Apart from price?
Price and TAM, AMD is not in Intel's position to experiment with overbuilt iGPUs that won't compete with dGPUs.
One more thing. If Ryzen 5 is both 6C and 4C lineups, that means that Ryzen 3 are going to be APUs.
Yes, Ryzen naming scheme even has a postfix for iGPU parts.
 

Glo.

Diamond Member
Apr 25, 2015
5,662
4,421
136
Price and TAM, AMD is not in Intel's position to experiment with overbuilt iGPUs that won't compete with dGPUs.
TAM for 4C/8T+ iGPU is much bigger than for 4C/8T only CPU.

AMD will sell much more APUs this year than 4C/8T CPUs, only.

P.S. Are you sure, that having technology advantage over Intel is not being in position to experiment with technology? ;)

Think about how much faster will be Vega with 1024 GCN cores, with access to HBM2 over Intel best offerings here ;).
 
  • Like
Reactions: Drazick
May 11, 2008
19,306
1,131
126
I was wondering about something. I read that for some situations like gaming, it is better to disable SMT.
But when SMT is disabled, the 8 threads on the cores can just as often be stalled as with SMT enabled but other threads take over execution time, filling in the gaps. Only, difference is that when SMT is disabled, when the thread stalls, the core stalls as well with SMT disabled. This would mean that Ryzen would dissipate less heat with SMT disabled because there are moments when the core does less work, allowing for higher sustained overclocks and boost clocks.

Does that make any sense or am i overlooking something ?

I mean, with 8 cores, that would be sufficient for a lot of people.
Also, i have read that going for 2666MHz single sided (ranked, only memory chips on 1 side ) DDR4 modules is the best option to get high memory speeds.

Is that true ?
 
  • Like
Reactions: french toast

french toast

Senior member
Feb 22, 2017
988
825
136
TAM for 4C/8T+ iGPU is much bigger than for 4C/8T only CPU.

AMD will sell much more APUs this year than 4C/8T CPUs, only.
Your right cost wise they could do it with expensive ultrabook apus, but im thinking of engineering hours, i think amd are fully booked right now with small budget.